From 7fd1b169f38fb5791ce3c6c675649b864dd4bcb6 Mon Sep 17 00:00:00 2001
From: kbkpbot <kbkpbot@gmail.com>
Date: Tue, 6 Feb 2024 23:53:02 +0800
Subject: [PATCH] builtin: add a string.u8_array() method (#20736)

---
 vlib/builtin/string.v      | 53 ++++++++++++++++++++++++++++++++++++++
 vlib/builtin/string_test.v | 46 +++++++++++++++++++++++++++++++++
 2 files changed, 99 insertions(+)

diff --git a/vlib/builtin/string.v b/vlib/builtin/string.v
index 24c388ec11df42..e0ccfa33b86a36 100644
--- a/vlib/builtin/string.v
+++ b/vlib/builtin/string.v
@@ -613,6 +613,59 @@ pub fn (s string) u8() u8 {
 	return u8(strconv.common_parse_uint(s, 0, 8, false, false) or { 0 })
 }
 
+// u8_array returns the value of the hex/bin string as u8 array.
+// hex string example: `'0x11223344ee'.u8_array() == [u8(0x11),0x22,0x33,0x44,0xee]`.
+// bin string example: `'0b1101_1101'.u8_array() == [u8(0xdd)]`.
+// underscore in the string will be stripped.
+pub fn (s string) u8_array() []u8 {
+	// strip underscore in the string
+	mut tmps := s.replace('_', '')
+	if tmps.len == 0 {
+		return []u8{}
+	}
+	tmps = tmps.to_lower()
+	if tmps.starts_with('0x') {
+		tmps = tmps[2..]
+		if tmps.len == 0 {
+			return []u8{}
+		}
+		// make sure every digit is valid hex digit
+		if !tmps.contains_only('0123456789abcdef') {
+			return []u8{}
+		}
+		// make sure tmps has even hex digits
+		if tmps.len % 2 == 1 {
+			tmps = '0' + tmps
+		}
+
+		mut ret := []u8{len: tmps.len / 2}
+		for i in 0 .. ret.len {
+			ret[i] = u8(tmps[2 * i..2 * i + 2].parse_uint(16, 8) or { 0 })
+		}
+		return ret
+	} else if tmps.starts_with('0b') {
+		tmps = tmps[2..]
+		if tmps.len == 0 {
+			return []u8{}
+		}
+		// make sure every digit is valid binary digit
+		if !tmps.contains_only('01') {
+			return []u8{}
+		}
+		// make sure tmps has multiple of 8 binary digits
+		if tmps.len % 8 != 0 {
+			tmps = '0'.repeat(8 - tmps.len % 8) + tmps
+		}
+
+		mut ret := []u8{len: tmps.len / 8}
+		for i in 0 .. ret.len {
+			ret[i] = u8(tmps[8 * i..8 * i + 8].parse_uint(2, 8) or { 0 })
+		}
+		return ret
+	}
+	return []u8{}
+}
+
 // u16 returns the value of the string as u16 `'1'.u16() == u16(1)`.
 @[inline]
 pub fn (s string) u16() u16 {
diff --git a/vlib/builtin/string_test.v b/vlib/builtin/string_test.v
index 4fee869f6c045e..4b5792bf5e6b71 100644
--- a/vlib/builtin/string_test.v
+++ b/vlib/builtin/string_test.v
@@ -811,6 +811,52 @@ fn test_to_num() {
 	assert big.i64() == 93993993939322
 }
 
+fn test_to_u8_array() {
+	// empty string
+	assert ''.u8_array() == []
+	assert '0x'.u8_array() == []
+	assert '0X'.u8_array() == []
+	assert '0b'.u8_array() == []
+	assert '0B'.u8_array() == []
+	// invalid digit
+	assert '-123'.u8_array() == []
+	assert '1_2xt'.u8_array() == []
+	assert 'd1_2xt'.u8_array() == []
+
+	// ---------------------------------
+	// hex test
+	// invalid hex digit
+	assert '0X-123'.u8_array() == []
+	assert '0O12'.u8_array() == []
+	// odd number of digits
+	assert '0x1'.u8_array() == [u8(0x01)]
+	assert '0x123'.u8_array() == [u8(0x01), 0x23]
+	assert '0x1_23'.u8_array() == [u8(0x01), 0x23]
+
+	// long digits
+	long_u8 := [u8(0x00), 0x11, 0x22, 0x33, 0x44, 0x55, 0x66, 0x77, 0x88, 0x99, 0xaa, 0xbb, 0xcc,
+		0xdd, 0xee, 0xff]
+	assert '0x00112233445566778899aabbccddeeff'.u8_array() == long_u8
+	assert '0x00_112_2334455667788_99aabbccddeeff'.u8_array() == long_u8
+	assert '0x00112233445566778899AABBCCDDEEFF'.u8_array() == long_u8
+	assert '0x001_12233445566778899A_ABBCCDDEEFF'.u8_array() == long_u8
+
+	// ---------------------------------
+	// bin test
+	// invalid bin digit
+	assert '0b-123'.u8_array() == []
+	assert '0B12'.u8_array() == []
+	// not enough length
+	assert '0b0'.u8_array() == [u8(0x00)]
+	assert '0b1'.u8_array() == [u8(0x01)]
+	assert '0b101'.u8_array() == [u8(0x05)]
+	assert '0b0101'.u8_array() == [u8(0x05)]
+	// long digits
+	assert '0b0101_0101'.u8_array() == [u8(0x55)]
+	assert '0b0101010110101010'.u8_array() == [u8(0x55), 0xaa]
+	assert '0b0101010110101010_0101010110101010'.u8_array() == [u8(0x55), 0xaa, 0x55, 0xaa]
+}
+
 fn test_inter_format_string() {
 	float_num := 1.52345
 	float_num_string := '-${float_num:.3f}-'